A&A 474, 873–882 (2007)DOI: 10.1051/0004-6361:20077540c© ESO 2007

Astronomy&

Astrophysics

A multiwavelength study of the S106 region

III. The S106 molecular cloud as part of the Cygnus X cloud complex

N. Schneider1,4, R. Simon4, S. Bontemps2, F. Comerón3, and F. Motte1

1 DAPNIA/SAp CEA/DSM, Laboratoire AIM CNRS - Université Paris Diderot, 91191 Gif-sur-Yvette, Francee-mail: nschneid@cea.fr

2 OASU/LAB-UMR5804, CNRS, Université Bordeaux 1, 33270 Floirac, France3 ESO, Karl-Schwarzschild Str. 2, 85748 Garching, Germany4 I. Physikalisches Institut, Universität zu Köln, Zülpicher Straße 77, 50937 Köln, Germany

Received 26 March 2007 / Accepted 15 August 2007

ABSTRACT

Context. The distance to the wellknown bipolar nebula S106 and its associated molecular cloud is highly uncertain. Values between 0.5and 2 kpc are given in the literature, favoring a view of S106 as an isolated object at a distance of 600 pc as part of the “Great CygnusRift”. However, there is evidence that S106 is physically associated with the Cygnus X complex at a distance of ∼1.7 kpc (Schneideret al. 2006, A&A, 458, 855). In this case, S106 is a more massive and more luminous star forming site than previously thought.Aims. We aim to understand the large-scale distribution of molecular gas in the S106 region, its possible association with other cloudsin the Cygnus X south region, and the impact of UV radiation on the gas. This will constrain the distance to S106.Methods. We employ a part of an extended 13CO and C18O 1 → 0 survey, performed with the FCRAO, and data from the MSX andSpitzer satellites to study the spatial distribution and correlation of molecular cloud/PDR interfaces in Cygnus X south. The 2MASSsurvey is used to obtain a stellar density map of the region.Results. We find evidence that several molecular clouds including S106 are directly shaped by the UV radiation from members ofseveral Cygnus OB clusters, mainly NGC 6913, and are thus located at a distance of ∼1.7 kpc in the Cygnus X complex. The definitionof OB associations in terms of spatial extent and stellar content in the Cygnus X south region is revised.

Key words. ISM: clouds – ISM: individual objects: S106 – ISM: molecules – ISM: kinematics and dynamics – radio lines: ISM

1. Introduction

The H II region S106 in Cygnus is famous for its bipolar emissionnebula (e.g., Oasa et al. 2006). The nebula is excited by the O-star S106 IR and embedded in a molecular cloud which was stud-ied in different molecular line tracers at various angular resolu-tions. Observations using lower-J CO lines at a resolution of ∼1′(Lucas et al. 1978; Bally & Scoville 1982; Schneider et al. 2002)revealed a cloud with an extent of ∼20 ′× 25′ and two emissionpeaks, separated by ∼3′, centering S106 IR. The eastern peak hasthe shape of a “lane” that is devoid of radio continuum emission,which is interpreted by some authors (Bally & Scoville 1982;Ghosh et al. 2003; Qin et al. 2005) as a massive disk or torusaround S106 IR. The western peak is also seen in NH3 emission(Stutzki et al. 1982). It is more diffuse and has an emission dis-tribution elongated north-south blocked by the lobes of the H II

region. A secondary site of star formation ∼5′ south of S106 IRwas detected as a peak in CO emission containing an embeddedcluster (Rayner 1994) seen also with 2MASS1 (Schneider et al.2002). See also the chapter “S106” by Hodapp & Schneider inthe “Handbook of star formation” (Reipurth 2007) for more in-formation on S106.

1 The Two Micron All Sky Survey (2MASS) is a joint projectof the University of Massachusetts and the Infrared Processing andAnalysis Center/California Institute of Technology, funded by theNational Aeronautics and Space Administration and the NationalScience Foundation.

The majority of the observations, however, concentrated onmapping the dense gas immediately around S106 IR at higherangular resolution in various atomic and molecular tracers (e.g.,Churchwell & Bieging 1982; Little et al. 1995; Schneider et al.2002, 2003) and the continuum (Richer et al. 1993; Vallée et al.2005; Motte et al. 2007). Optically thick molecular lines re-vealed strong outflow emission close to S106 IR. This star alsopowers a bright photon dominated region (PDR) as seen inatomic Far-infrared (FIR) lines ([C II] 158 µm, [C I] 809 and370 µm, Schneider et al. 2003) and emission lines of Brγ andH2 1–0 S(1) (Hayashi et al. 1990).

The range of distances given in the literature for the S106complex is broad. Early estimates using the kinematic informa-tion of S106 give values >2 kpc (e.g., Reifenstein et al. 1970;Maucherat 1975). However, S106 is located at Galactic longi-tude ∼76.5◦ where the local Galactic arm, the Perseus arm, andthe outer Galaxy are found along the same line of sight, cov-ering distances between a few hundred pc and 8 kpc. In thisdirection, radial velocities of nearby clouds are close to zeroand thus do not provide reliable distance estimates. Distancesderived with other methods (Eiroa 1979; Staude 1982; Rayner1994) are smaller (500 to 600 pc). These values are commonlyused for the S106 region.

In this paper, we study the spatial and kinematic distribu-tion of molecular gas in S106 and place this cloud in the con-text of other objects in the Cygnus X region. To date, extrin-sic influences on S106 (shock-compression, radiation pressure,and winds from external OB-stars) have not been considered or

Article published by EDP Sciences and available at http://www.aanda.org or http://dx.doi.org/10.1051/0004-6361:20077540

874 N. Schneider et al.: S106. III.

studied because S106 was seen as an isolated object in the quies-cent “Cygnus Rift” at ∼600 pc distance. We investigate whetherthe UV radiation of the Cygnus OB1 and OB9 associations andtheir sub-associations, such as NGC 6913 (M29) and Ber87, af-fect the S106 molecular cloud and other molecular gas in this re-gion. For this, we compare 13CO emission line maps taken withthe FCRAO2 with mid-IR emission data from MSX3. Data from2MASS are used to determine stellar densities of the brighteststars in order to obtain an overview of the distribution of the starclusters that influence the molecular gas. It is important to con-strain the distance to S106 since a larger distance implies thatthe molecular cloud is more massive and the exciting star S106IR has an earlier spectral type than previously thought and thuspowers a more luminous PDR.

Observational details are given in Sect. 2. Large- and small-scale overlays of 13CO onto mid-IR MSX data are shown inSect. 3, the observed molecular line maps are presented andsome cloud properties are determined from the CO data. In thediscussion (Sect. 4), we briefly review existing distance esti-mates for S106, discuss the “Cygnus Rift”, and finally proposea new scenario for S106, which is, in our view, closely associ-ated with the Cygnus X molecular cloud complex at a distanceof ∼1.7 kpc (Schneider et al. 2006). Section 5 summarizes thepaper.

2. Observations

2.1. FCRAO CO observations

CoverageWe use data from an extensive molecular line survey of thewhole Cygnus X region taken with the FCRAO 14m telescope.An area of 35 deg2 was observed in 13CO and C18O 1 → 0using the 32 pixel “Second Quabbin Optical Imaging Array”(SEQUOIA) in an On-the-Fly (OTF) observing mode. The datawere obtained between 2003 December and 2006 January andwill be presented in more detail in Simon et al. (2007, in prep.).In the current paper, we concentrate on 13CO 1 → 0 emissionfrom an area of 3 × 3 deg2 covering the molecular clouds in thesouth-western part of the survey (named “Cygnus X south” inSchneider et al. 2006), and in 13CO and C18O 1 → 0 emissionfrom a region of 40′ × 40′ extent, centered on S106.OTF-mappingThe Cygnus X data were assembled from individual OTF maps(footprints) of 20′ × 10′ size. The integration time was 1 s foreach of the 64 simultaneously recorded spectra, 32 in 13CO and32 in C18O, as the telescope moved continuously across the sky.One OTF footprint map thus took ∼30 min including overheads.The maps were observed with some overlap to compensate forthe less dense sampling at the edges of the footprint maps andallow for uniform noise in the final, merged data set. Due to theOTF observing mode, the resulting grid of spectra is very denseand highly oversampled. The spectra were gridded on a 22.′′5raster before being merged into the final data set. The FWHMof the FCRAO at the observed frequencies is ∼46′′. The spectrahave a mean 1σ rms noise level of ∼0.2 K on a T ∗A antenna tem-perature scale, i.e., not corrected for the main beam efficiency of∼0.48 at 110 GHz.BackendThe receiver was used in combination with a dual chan-nel correlator (DCC), configured to a bandwidth of 25 MHz,

2 Five College Radio Astronomy Observatory.3 Midcourse Space Experiment.

1024 channels, and a velocity sampling of 0.066 km s−1. Withthis correlator, two independent intermediate frequencies (IFs)could be processed for each of the 32 pixels, and therefore, 32spectra could be observed simultaneously at 110 and 109 GHz(13CO and C18O).Pointing and CalibrationPointing and calibration were checked regularly at the start ofthe Cygnus LST interval and after transit of Cygnus (no obser-vations were performed at elevations higher than 75◦). Pointingsources were SiO masers of evolved stars, i.e., χ-Cyg, R-Leoand T-Cep, depending on LST-time. The calibration was checkedregularly on the position of peak emission in DR21 and found tobe consistent within 10%.

3. Results and analysis

3.1. Large-scale correlation of molecular lineand mid-IR emission

Figure 1 shows a part of the 13CO 1→0 FCRAO survey of theCygnus X region overlaid as contours on a color image of mid-IR emission at 8.3 µm (MSX Band A with an angular resolutionof 20′′). The CO emission is integrated over a velocity range of–10 to 20 km s−1 so that all molecular clouds seen in Cygnus X(Schneider et al. 2006) are included. The molecular cloud asso-ciated with S106 is located at l = 76.4◦ and b = −0.7◦.

The emission at 8 µm is a tracer of PDR-interfaces becausethe UV radiation from massive stars heats small grains and poly-cyclic aromatic hydrocarbons (PAHs) which re-radiate part oftheir energy at this wavelength. The point-like sources are either(clusters of) young massive and bright stars or embedded redobjects that are usually massive protostars. The most prominentfeature seen in the MSX image is a large-scale circle (indicatedby a long-dashed line in Fig. 1) of bright mid-IR emission ap-proximately centered on NGC 6913. The circle is not closed,but shows a lack of emission towards the cluster Ber86 and thecenter of Cyg OB1. The star-forming regions within the molec-ular clouds are found to be arranged on this circle. Some containidentified star clusters (Cl02–5, Cl07 after the nomenclature ofLe Duigou & Knödlseder 2002; and Ber87), others show strongmid-IR emission due to embedded clusters (e.g., at l = 76.2◦,b = 0.1◦) and/or due to external excitation. The most prominentexamples for the latter case are found at l = 76.8◦, b = 2.1◦(Cloud A), at l = 77.5◦, b = 1.8◦ (Cloud B), and S106. They arebright-rimmed, globular-shaped molecular clouds with a tail ofemission in mid-IR and CO pointing away from the circle centerand are discussed in more detail in the next section.

There are, however, two examples where CO emission andmid-IR emission do not correspond well, i.e., towards an ex-tended, diffuse area south-east and east of NGC 6913, and a re-gion of rather compact clouds south of Cl07. The non-correlationwith mid-IR emission does not imply that these clouds are notcorrelated with the Cygnus X clouds since they can be slightlydisplaced in the fore- or background with respect to the bulkemission. However, they can also be related to the “CygnusRift”. We will come back to this point in Sect. 4.2.

The large scale distribution of mid-IR/CO emission suggeststhat the center of excitation for the PDR cloud surfaces is locatedclose to the circle center. Obviously, the open cluster NGC 6913is the first candidate to be the exciting source. It is seen as oneof the nuclei of OB1 and contains around 100 stars (Wang &Hu 2000) even though only 2 OB-stars were identified (Wang &Hu 2000; Boeche et al. 2004). Other nuclei of OB1 are Ber86and Ber87 close to NGC 6913. At this point, we only state that a

N. Schneider et al.: S106. III. 875

Fig. 1. Overlay of 13CO 1 → 0 emission inthe velocity range –10 to 20 km s−1 (whitecontours) on an 8 µm image of MSX in theCygnus X south region containing S106. Themap shows about 25% of the FCRAO surveyarea (Simon et al. 2007, in prep.) the bound-ary of which is indicated by a short-dashedline. Contours are from 4 to 36 K km s−1 insteps of 4 K km s−1, the color scale is (0.05–1.0)× 10−5 W m−2 sr−1. The large black ellipsesrepresent the extent of the OB8, OB9 andOB1 associations and crosses are their cen-ter. Yellow, small circles show known clus-ters, i.e., NGC 6913, NGC 6910, Ber86, Ber87,as well as Cl02–7 (Le Duigou & Knödlseder2002). Small circles with numbers indicateknown H II regions DR4, 5, 6, 9, 12, and 13(Downes & Rinehart 1966). The long-dashedcircle encloses a cavity largely devoid of mid-IR emission.

source of UV radiation close to the location of NGC 6913 mayhave created the observed cavity in mid-IR emission and the cir-cle of PDR emission where the ionization fronts hit the (molec-ular) gas.

3.2. Small-scale correlation of molecular lineand mid-IR emission

Figure 2 shows the prominent PDR interfaces of S106 and twoother clouds in detail (Cloud A and B in Fig. 1) to illustrate theeffect of stellar winds/UV radiation that are shaping the molec-ular clouds. In addition, we show position-velocity cuts in 13COto illustrate the velocity structure.

In Cloud A (Fig. 2a), two mid-IR point sources are located atthe tip (∼2 pc separation, assuming a distance of 1.7 kpc) of thedense molecular cloud. Around these sources, a ring-like struc-ture in mid-IR emission is observed that corresponds to VdB130,a single star or small star cluster in a reflection nebula, similarto “Cloud B” (see below). The H II region DWB34 listed in thecatalog of Dickel et al. (1969) is indicated in the plot. The tail(∼10 pc length) of the cloud is more dispersed in CO and mid-IR emission and cleary points away from the cavity (comparewith Fig. 1). The position-velocity cuts (lv-cuts) go along thetwo main flow directions of the gas. Only a small velocity gradi-ent can be identified for Cut 2. Cut 1, however, shows a strong,clearly defined gradient in velocity (+0.5 km s−1/pc). Similarvalues are found in much smaller molecular “globules” (sizesof typically 0.5 pc for the head and 1.5 pc for the tail, see, e.g.,White et al. 1997). However, the mechanisms at work that are re-sponsible for the observed large-scale gas flows in Cloud A aremost likely the same, i.e., influence of UV-radiation and stellarwinds of OB stars.

The structure of mid-IR and 13CO emission around Cloud B(Fig. 2b) is very complex and we observe several, overlayingfeatures. The large scale emission in both tracers shows a simi-lar distribution as for Cloud A with a dense clump of moleculargas, facing the UV field, and a more dispersed tail. Embedded in

the main CO peak, an MSX point source corresponds to Mol121(Molinari et al. 1998). This is an ultra-compact H II region whichcontains a small cluster of stars seen in near-IR by Deharvenget al. (2005). The same authors identified a nearly circular H II

region (their source “X” and marked as such in the plot) thatwas not listed in the DWB-catalog of optically visible H II re-gions. The edge of the main CO peak is lit up by UV radiation,as seen in the bright, narrow rim of MSX emission. Extendingto the right and left of the CO peak, a bow-shaped feature in COand mid-IR emission is seen, suggesting that here we also ob-serve compressed, UV illuminated, dense gas. This structure isclearly visible in the lv-cut 2 (between 7′ and ∼10′), covering avelocity range of ∼6 km s−1 that is much larger than a typical in-ternal velocity dispersion of a quiescent molecular cloud. Cut 1goes along one of the main tail axes. The outflow from Mol121is seen at ∼8′. The CO emission in the tail is complex, suggest-ing on first sight a second outflow source around 18′. However,a careful analysis of the spectra shows that the large range ofvelocities observed (–3 to 4 km s−1) is mainly caused by severaloverlapping cloud components and a high internal velocity dis-persion of the gas. The existence of an outflow source can not beruled out, but requires higher angular resolution molecular lineobservations.

In Cloud A and B, the mid-IR emission is clearly shiftedtowards the edge of the clouds and the center of the cavity. Thisstratified distribution of CO and mid-IR emission together withthe globular shape of the clouds is strong evidence that the originof the exciting UV photons is inside the large-scale circle.

The lower intensity mid-IR emission in S106 forms a bow-shaped feature with two diffuse tails pointing away from the cen-ter of the circle in Fig. 1. This shape is clearly caused by ex-ternal UV radiation from inside the large-scale cavity. The COemission has a similar distribution but is more concentrated tothe inside of the bow-shaped mid-IR emission, suggesting thatthe molecular cloud is embedded in a larger cocoon of heatedgas, visible in mid-IR emission. The high-intensity mid-IR emis-sion is the well-known bipolar nebula S106 whose orientation is

876 N. Schneider et al.: S106. III.

Fig. 2. Top: cloud A a), B b) and S106 c) in 13CO 1 → 0 emission (contours) overlaid on mid-IR emission (8 µm from MSX). The bold linesindicate two position-velocity cuts shown in the bottom figures. An offset of 0′ indicates the lower latitude starting point of the cuts. Particularsources are labeled.

nearly orthogonal to the low-intensity mid-IR emission. The ex-citing UV source is the OB-star S106 IR. Though this star cre-ates a bright PDR in its immediate environment, the UV fluxdecreases rapidly from a few 105 G◦ at the position of S106 IRdown to 102 G◦ in 2′ distance (Schneider et al. 2003). Cut 1and 2 show a velocity gradient that is smaller (∼0.3 km s−1/pc)than for Cloud A. It seems that the gas here is less dynamic thanalong Cut 1 in Cloud A and in the shock-compressed region ofCloud B. Strongly visible in both cuts is the outflow emissioncaused by S106 IR.

In the next section, we discuss in more detail the molecularline emission of the prominent source S106.

3.3. Molecular line maps of S106

The results of our molecular line mapping of S106 in the 13COand C18O 1 → 0 transitions are shown in Figs. 3 and 4. InFig. 3, the 13CO map is overlaid on a color image of 4.5 µm

emission obtained from the Spitzer space telescope archive.Figure 4 shows the line integrated 13CO and C18O 1 → 0 mapsseparately.

Significant 13CO emission is found in the velocity rangev = −9 to 6 km s−1 while the C18O 1 → 0 emission is restrictedto the velocity range v = −4 to 2 km s−1. Both line maps showthe characteristic two-lobe emission distribution east and westof S106 IR (indicated as a star in the plot). The lobes of ionizedgas from the H II region are nearly orthogonal to the CO emis-sion. This morphology is due to the impact of the radiation andthe ionizing wind of S106 IR that created the bipolar lobes whichare probably collimated by a very small disk (Persson et al. 1988;Bally et al. 1998). The very turbulent layer between ionized andmolecular gas in the lobes becomes visible in Hα images takenwith the Hubble Space Telescope (Bally et al. 1998). Beyondthese lobes, the cooler molecular gas wraps around the H II re-gion, visible as the bulk emission of the cloud between –3 and3 km s−1. The strong CO peak southeast of S106 IR is due to

N. Schneider et al.: S106. III. 877

Fig. 3. Line integrated (–9 to 6 km s−1) 13CO 1 → 0 map overlaid ona color picture of Spitzer 4.5 µm data. CO-contours go from 2 (3σ) to46 K km s−1 in steps of 4 K km s−1 and the IR-color range from 0 to1.6 × 106 MJy/sr. The exiting star S106 IR is marked by a star and apeak of NH3 emission by a triangle. Several secondary star formationsites are indicated, including S106 south and south-east.

outflow emission where the molecular gas is swept up from thecavity walls. It is this part of the S106 molecular cloud that wasfrequently observed in molecular lines and discussed so we donot go into more detail here. See Schneider et al. (2002) and ref-erences therein, as well as a recent publication by Vallée & Fiege(2005).

The 13CO J = 1 → 0 line traces rather low gas densitiesaround 3 × 103 cm−3 and thus gives an overview of the total gasdistribution of the S106 molecular cloud. The emission extendsmore than 20′ to the south, separated into two tails each showinga (weak) peak of emission at its end. The average intensity is afactor ∼3 lower than for the two regions of peak emission closeto S106 IR. In contrast, the C18O map shows almost no emis-sion features from the two tails but traces on the more shielded,denser core of molecular gas.

Higher-density tracers such as CS 2 → 1 and N2H+ 1 → 0show the same emission distribution (Schneider et al. 2007, inprep.). The 13CO (C18O) 1 → 0 main beam brightness temper-atures vary between a few K (<1 K) in the halo region and thetwo cloud tails and rise up to ∼15 K (∼4 K) in the bright easternand western lobes close to S106 IR.

The Spitzer 4.5 µm data reveal more clearly than the MSXimage (Fig. 2) the secondary star formation sites in S106. TheS106 south cluster, first deteced by Rayner (1994) and seen in2MASS images (Schneider et al. 2002), is still embedded in thebulk emission of 13CO and is located at a peak of C18O emission.

Fig. 4. Maps of the line integrated 13CO 1 → 0 (top) and C18O 1 → 0(bottom) emission in the velocity range v = −9 to 6 km s−1 and v = −4 to2 km s−1, respectively. Contours go from 2 (3σ) to 46 K km s−1 in stepsof 4 K km s−1 for 13CO and 2 (3σ) to 8 K km s−1 in steps of 1 K km s−1

for C18O.

At l = 76.4◦, b = −0.83◦ and at l = 76.35◦, b = −0.68◦ ratherstrong sources are detected but since we do not observe extendedemission around these objects, they may well be foregroundstars. In contrast, the source in between shows extended emis-sion and may constitute a small cluster of stars. Conforming tothe nomenclature in Schneider et al. (2002), we call this source“S106 south-east”. It is, however, not associated with a promi-nent peak in CO emission. All objects are too far separated inprojected distance to be attributed to the cluster of fainter starsdetected by Hodapp & Rayner (1991) which has a 1.7′ radius forthe 1σ-level of the stellar density distribution.

3.4. Physical properties of the S106 molecular cloud

We decomposed the 13CO 1 → 0 map into clumps with aGaussian density and velocity distribution, using the algorithmGaussclumps. See Stutzki & Güsten (1990) and Kramer et al.(1998) for more details on the program and, e.g., Schneider& Brooks (2004) and Simon et al. (2001) for its application.The clump masses were then calculated assuming opticallythin emission and LTE (local thermodynamic equilibrium) as

878 N. Schneider et al.: S106. III.

outlined in Schneider et al. (1998). We assume a distance of1.7 kpc for our mass calculation (see Sect. 4). The total massof the cloud is then determined to be the sum of all decomposedclumps (98% of the fitted total intensity, 581 clumps in total)which is 7620 M�. The average density within the clumps isthen 103 cm−3. The mass is consistent with former estimationswhen scaled up, with a distance of 1.7 kpc. Bally & Scoville(1982) determined 9710 M� based on FCRAO 13CO 1→ 0 data,and Lucas et al. (1978) estimated a mass of 12 000 M� using13CO 1→ 0 observations from the MWO 5 m telescope.

A fit to the clump-mass distribution (the number of clumpsdN within the mass interval dM, described by the power-lawdN/dM ∝ M−α, see Kramer et al. 1998) gives α = 1.62. Thisis in accordance with values around 1.7 determined from for-mer CO observations at different angular scales (Schneider et al.2002).

4. Discussion

4.1. Distance estimates for S106

The commonly used distance for S106 is <1 kpc. Earlykinematic distances gave much larger values, i.e., 2.3 kpc(Reifenstein et al. 1970; Pipher et al. 1976), and 5.7 kpc(Maucherat 1975). Since local clouds in the direction of S106have radial velocities close to zero (see Sect. 1), kinematic dis-tances are not reliable. Eiroa et al. (1979) determined a distanceof 500 pc with IR-photometry of the exciting star S106 IR. Thismethod, however, requires a source model for S106 IR and doesnot give a single solution. In their paper, the best fit for the en-ergy spectrum of S106 was obtained with a B0–O8 main se-quence star at Av = 21m. Staude et al. (1982) derived a distanceof 600 pc using UBVI photometry of 8 stars which already rep-resents a very low number statistic. They observed a sudden in-crease by two magnitudes of Av starting at a distance of 600 pcand attribute that to extinction due to the “Cygnus Rift”, an ex-tended area of gas at that distance. This, however, does not implythat the molecular cloud of S106 is located at the distance of therift; it could well be further away. Due to these uncertainties, thedistances derived by the methods outlined above were long dis-puted. Neckel (1990, priv. communication) favors a distance of2 kpc on the basis of a larger number of foreground stars seenon Palomar Observatory Sky Survey Plates. Rayner (1994) de-termines a distance of at least 1.2 kpc by using J–H versus H–Kdiagrams, equally based on a large number of star counts.

Another argument to place S106 at a distance of ≤600 pccould be that the H II region is partly visible in the optical andmust thus be located in front of or within the “Cygnus Rift”.However, in the following section, we give arguments againstthis scenario.

4.2. S106 and the “Cygnus Rift”

The “Cygnus Rift” is a large area of obscuration in optical im-ages of the Cygnus X region and is probably part of the “GouldBelt”, an extended (∼700 × 1000 pc) cloud structure of low-density gas inclined 20◦ to the Galactic plane (Guillot 2001).An Hα image and extinction map presented in Schneider et al.(2006, their Fig. 1) illustrate the large-scale distribution of thisfeature. The distance to the Cygnus Rift was determined by pho-tometric studies, concluding that the first reddened stars appearat 430 pc (Straizys et al. 1993; North America Nebula at l = 84◦)and 800 pc (Ikhsanov 1961, IC 1318 b/c at l = 78◦). TowardsS106, Staude et al. (1982) noticed a jump in Av of around 600 pc.

Figure 5 shows the southern part of the Hα image and an extinc-tion map with Av up to 10m (the maximum value is Av = 32m)with contours of IRAS 100 µm emission overlaid on both im-ages. Comparing both plots, it becomes obvious that the overalllarge-scale extinction is low and varies across the region. TheH II region IC 1318 b/c is clearly visible in the optical imagethough its distance is around 1.5 kpc (see Schneider et al. 2006and references therein). The visual extinction in this direction isa few magnitudes (<5m). S106 appears in the optical image evenmore obscured than IC 1318 b/c. Its extinction averaged over the2′ 2MASS resolution is 11m. Both IC 1318 and S106 are locatedat the edge of the large-scale smooth, yet low level, extinction ofthe Cygnus Rift.

However, since one has to distinguish between the diffuse ex-tinction due to the rift and the extinction due to dust within themolecular clouds of Cygnus X (associated with the Cyg OB2 as-sociation at a distance of 1.7 kpc) or clouds located further away,we overlaid channel maps of 12CO 1 → 0 emission taken fromthe CfA CO survey (Dame et al. 1987) on the optical image. Theonly velocity range matching approximately the region of obscu-ration is found between 6 and 20 km s−1. This component wasalready tentatively identified with the rift emission by Schneideret al. (2006) and Piepenbrink & Wendker (1988). Overlays withour FCRAO 13CO 1 → 0 data did not conclusively reveal therift, which implies that the column density of the molecular gasin the rift is very low so that the 13CO line emission is weak. Thisinterpretation is supported by the 100 µm emission that is due tocold dust mixed with molecular gas. As can be seen in Fig. 5b,the 100 µm emission traces much more clearly the dense molec-ular clouds than the diffuse rift emission. That again implies alow column density of the absorbing material in the rift.

To better separate the extinction due to molecular clouds andthe rift, we arbitrarily distinguish the two features by a (blue)Av = 5m contour line. (Though one has to be aware that the riftextinction also covers part of the Cygnus X clouds). This value isjustified since the rift extinction is estimated to be not more thana few magnitudes (Dickel & Wendker 1978). However, not allemission within the Av > 5m contour, including S106, necessar-ily arises from one coherent cloud complex at the same distance.The situation is much more complex due to possible backgroundfeatures. For example, Boeche et al. (2004) noticed that extinc-tion towards NGC 6913 occults part of the cluster and assignedthat to a “thick interstellar cloud” in the foreground. Figure 5shows that the obscuration of NGC 6913 is in fact partly dueto the low-density Cygnus Rift but we also detected significant12CO and 13CO 1 → 0 emission (Fig. 1) in that direction andin direction Berkely 87 at velocities between –10 and 6 km s−1.These emission features are not prominent in mid-IR emission,implying that they are possibly background clouds and not asso-ciated with Cygnus X.

We suggest now that the high extinction values observed to-wards S106 are due to its molecular cloud, which is linked toother cloud complexes in the Cygnus X south region throughan observed common influence of UV radiation on the clouds.Therefore, S106 is not an isolated object in front of or withinthe Cgynus Rift, but it is part of the large-scale cloud complexesof the Cygnus X region at the distance of the OB clusters, i.e.,larger than 1 kpc.

4.3. S106 as part of the Cygnus X star-forming region

In Sect. 3.1, we presented clear evidence that the UV radiationof the Cygnus OB1 association and its sub-associations such asNGC 6913 (M29) and Ber86 affect the S106 molecular cloud

N. Schneider et al.: S106. III. 879

Fig. 5. a) Inverted Hα image (courtesy of William McLaughlin) of theCygnus X south region. Overlaid are white contours of IRAS 100 µmemission (levels 50, 300, 500, 2000, 4000, 5000 MJy/sr) and red con-tours of 12CO 1→ 0 emission (levels 0.5, 1.5, 3, 5, 7, 9 K km s−1 in thevelocity range 6 to 20 km s−1) taken from the Center for AstrophysicsCO-survey (Dame et al. 1987). b) Extinction map of the same region ingrey scale (Av = 0 to 10m). This map was obtained from the reddeningof J −H and H −K colors using 2MASS data (Bontemps et al. 2007, inprep.). The Av = 5m contour is outlined in bold blue. Overlaid are thesame white contours of IRAS 100 µm emission. The dashed light bluepolygon indicates the H II region IC 1318 b/c. Known star clusters areindicated in yellow in both plots.

and at least two other regions (Cloud A and B). The distanceto Cyg OB1 is determined to be 1.25–1.83 kpc (from a compi-lation of several references in Uyaniker et al. 2001). The dis-tance to NGC 6913 is quoted as between 1.1 and 2.4 kpc (seeBoeche et al. 2004, and references therein). The three molecu-lar cloud regions discussed in detail in this paper are part of aring-like structure (or a “bubble” if extrapolated to 3D) whichis prominent in mid-IR and molecular line emission and showsseveral active sites of star formation (Clusters Cl2–5 and 7, DR5,

9, and 13). Clouds A and B are linked to the molecular cloudcomplex of “Group IV” in the “Cygnus X south” region (classi-fication from Schneider et al. 2006). It has been noted that thismassive (420 000 M�), homogeneous complex with emission be-tween –5 and 3 km s−1 is influenced by the Cyg OB2 cluster fromthe eastern side and the Cyg OB1/OB9 associations from thewestern side (in equatorial coordinates), and that the molecularclouds are located in between. The bulk emission from the S106molecular cloud is seen in the same velocity range and togetherwith the fact that S106 is shaped by UV radiation and is foundon a circle of star-forming sites, we see S106 now as a part ofGroup IV at the same distance of ∼1.7 kpc. This value, however,remains approximate since all clouds in Cygnus X were giventhis common distance due to their correlation with Cyg OB2.Moreover, S106 lies on a “bubble” created by Cyg sub-OB1 as-sociation, thus its distance must be within the distance limits ofthis cluster (1.25–1.83 kpc).

What remains now is to identify more clearly the majorsource(s) of UV radiation/stellar winds which shape the molec-ular clouds on the star-forming circle. For that, we investigatein the following section the properties of the Cyg OB1 clus-ter and its sub-clusters as well as the general distribution of theCygnus X OB associations.

4.4. S106 and the Cygnus OB associations

Originally, OB associations were recognized from catalogs ofhot stars observed spectroscopically in the optical (Blaauw 1964;Ruprecht et al. 1981; Humphreys 1978). The distances to thesestars can be derived (though with large uncertainties) from thecalibration of absolute magnitudes with spectral types on themain sequence. In the direction of Cygnus X, 9 OB associationshave been defined by Humphreys (1978). These early works arevery sensitive to extinction since obscured stars could not beincluded. The large uncertainties on the individual distances toeach star hamper any attempt to get precise distances for groupsof stars. In addition, the distance uncertainties make it difficultto recognize coherent groups, especially in crowded regions suchas Cygnus.

The distribution of bright 2MASS sources now provides away to recognize compact (and therefore young) clusters ofOB stars. The extinction is less in the near-IR, and from the near-IR colors (J − H and H − K) of the stars, one can roughly esti-mate the visual extinction toward each star and therefore correctthe observed magnitudes to recognize the intrinsically brighteststars. We have derived the stellar density of the brightest K-bandsources (magnitude brighter than 10m) using this extinction cor-rection (see Bontemps et al. 2007, in prep., for details). The re-sulting contour map overlaid on the MSX image is displayed inFig. 6. The positions and sizes of the 4 OB associations in theS106 region from the catalog of Uyaniker et al. (2001) are indi-cated as well. A comparison with the contours of stellar densityshows that Cyg OB2 at l = 80◦ is the richest and best definedassociation and corresponds to the former definition. Towardslower longitudes along the Galactic plane (b ∼ 0.5◦), the stellardensities are still high but less consolidated. Within the ellipsecharacterizing Cyg OB9, a rather compact nucleus can be iden-tified. Towards NGC 6913, two peak regions appear which wename sub1-OB1/9 and sub2-OB1/9. They are not identical withthe center of the Cyg OB1 and/or OB9 associations but ratherform sub-clusters. Inside the Cyg OB1 region, Ber87 shows adetectable cluster of bright stars but in addition, several higherstellar density regions are found.

880 N. Schneider et al.: S106. III.

Fig. 6. Contours of stellar densities for starsbrighter than AK = 10m overlaid on a grey scaleplot of mid-IR emission taken with MSX. Red,bold contours are 1.1, 1.2, 1.3, the green con-tour is 0.9, and the blue contour 0.7. The ex-tent of the Cyg OB 1, 2, 8 and 9 associationsare marked as dashed ellipses and known stellarclusters are indicated in the plot.

Mel’nik & Efremov (1995) proposed that OB1, OB8, andOB9 could actually be a single association centered at l = 76.8◦,b = 1.4◦ at a distance of 1.4 kpc. Inside these associations,several nuclei have been recognized: Ber 87, NGC 6913, andIC 4996 for Cyg OB1, and NGC 6910 for Cyg OB9. Altogether,however, only a few tens of optical hot stars are known (119 starsin OB1-8-9 in Mel’nik & Efremov 1995).

The full scenario is probably more complicated and will bediscussed in more detail in an upcoming paper (Bontemps et al.2007, in prep.). However, it becomes obvious that the defini-tion/classification of OB associations in the Cygnus X regionneeds to be revised. In any case, we can identify sub1-OB1/9 andsub2-OB1/9 (NGC 6913) as the main center of UV radiation forCloud A, B, and S106. The common effect of both sub-clusters ismost likely creating the cavity in mid-IR emission. It is clearlyvisible that the highest levels of stellar density fill up the bub-ble devoid of molecular gas. The circle of compressed gas aswe see it in CO and mid-IR emission (Fig. 1) marks the maxi-mum extent of the ionization front created by the sub-clusters.If we consider an expanding ionization front with a velocity of10 km s−1 and assume a radius of 30 pc (approximately what wesee around NGC 6913), the age of the cluster must be 3×106 yr.In the literature, we find values between 0.3–1.75 × 106 yr and10 × 106 yr (see Boeche et al. 2004, and references therein). Anolder age would fit in the scenario that NGC 6913 is not verycompact but already dispersed.

4.5. The central star S106 IR

The existence of a bright, highly reddened point source near thecenter of the S106 dark lane, identified as the ionizing source

and generally referred to as S106 IR, has been known since thefirst infrared studies of the region (Sibille et al. 1975; Allen &Penston 1975). Based on visible near-IR and broad-band pho-tometry, Eiroa et al. (1979) derived for S106 IR an approximatedistance of 600 pc and a spectral type between B0 and O8. Thisdistance has been adopted in most subsequent studies, but it isimportant to keep in mind that it was indirectly derived usingstellar data, and ionization fluxes, based on estimates which havebeen superseded by further observations and models. The sameapplies to the estimate by Staude et al. (1982) which relies onthe detection of a relatively small (AV ∼ 2.5m) extinction jumpat d ∼ 600 pc (most probably the well known Cygnus rift) whichdoes not prove that the S106 cloud is associated with it.

It thus seems timely to reassess the determination of the dis-tance to the ionizing star responsible for the ionization of S106using the results from state-of-the-art models. We use the in-frared photometry in the 2MASS catalog, which is largely freefrom the uncertainties due to resolution and aperture correctionsin the early literature based on single-element observations. The2MASS colors of S106 IR, (J −H) = 1.54, (H −KS) = 1.08, arenormal for an early stellar photosphere significantly reddenedby foreground extinction. Using the extinction law of Rieke &Lebofsky (1985) and the intrinsic infrared colors of mid- to late-O type stars, we derive AK = 1.86m, approximately correspond-ing to AV = 16.6m.

Visible spectroscopy of the nebula, which can be used toconstrain the temperature of S106 IR, has been presented bySolf (1980) and Staude et al. (1982). These latter authors mea-sured the [S II] λ6731 / [S II] λ6717 ratio indicative of a high den-sity (∼104 cm−3) consistent with the compactness of the nebula,whereas the [O III] λ5007/Hβ ratio corresponds to a temperature

N. Schneider et al.: S106. III. 881

in the range 34 000 K < T < 36 000 K according to the singleionizing star models of Stasinska & Schaerer (1997) based onthe CoStar stellar atmosphere models of Schaerer et al. (1996).The spectral types yielded by those models are in general be-tween one and two spectral subtypes later than those obtainedwith more recent atmosphere modeling (Martins et al. 2005).Based on the latter, the temperature interval given above roughlytranslates into a spectral type O8.5-O7.

Independent support for a temperature near the upper endof the estimated range has been reported by van den Anckeret al. (2000) based on mid-infrared spectroscopy obtained byISO, from which these authors derive a temperature of 37 000 K,corresponding to a spectral type O7 in the Martins et al. (2005)calibration. On the other hand, the main-sequence absolute mag-nitudes MV given by Martins et al. are relying on an empiricalcalibration based on emerged O stars, which produces brightermagnitudes than the zero-age main sequence (ZAMS) CoStarmodels. We have thus used the CoStar temperature-absolutemagnitude relationship as being more adequate for an object sus-pected to be very near the ZAMS such as S106 IR. In this way,we estimate −3.4 > MV > −3.8 for S106 IR, or −2.57 > MK >−2.90 using the intrinsic (V − K) colors of O stars from Martins& Plez (2006). Used in combination with KS = 9.49 measuredby 2MASS and the estimated extinction listed above, we obtaina distance modulus 10.20 < DM < 10.53, with a preference ofthe upper limit based on the ISO spectroscopy. The correspond-ing distance lies between 1.1 kpc and 1.3 kpc, somewhat belowour proposed distance of 1.7 kpc (DM = 11.15) for the S106region but clearly above the 600 pc (DM = 8.9) proposed inearlier studies.

An earlier spectral type for S106 IR implies that moreUV photons are produced, creating a more intense UV fieldwhich is responsible for the illumination of a bright PDR.Supposing S106 IR is an O8 star, a UV field of 1.5 × 106 G◦at a distance of 5′′ = 0.04 pc or 1.3 × 105 G◦ at a distance of17′′ = 0.13 pc is generated. These values are consistent withthe results for the UV field close to S106 IR obtained from ob-servations of FIR-PDR lines (Schneider et al. 2003). One canalso directly estimate the ionising flux of S106 IR from the to-tal radio flux of the H II region. We took the flux at 4.8 GHzfrom Wendker et al. (1991), of 13.02 Jy which corresponds, ford = 1.7 kpc, to an emission measure EMV of 1.29×1061 cm−3 inthe optically thin case. Using the typical recombination rate of2.7×10−13 cm3/s (Spitzer 1978), this EMV requires a Lyman fluxof 3.5 × 1048 s−1. On the ZAMS, this ionising flux correspondsto an O7 star (Meynet et al. 1994) which is fully consistent withthe above estimates of the spectral type of S106 IR.

Though we are confident in our new distance estimate forS106, the ultimate answer could be given by VLBI parallaxmeasurements of the H2O maser in S106 FIR. Furuya et al.(1999) observed this maser using VLBA during 4 monthly peri-ods which can already serve as a starting point. The future ESAastrometric space mission GAIA will also give the parallax of alarge number of at least the OB stars of the region.

5. Summary

By comparing the emission distribution of the 13CO 1→ 0 line,observed with the FCRAO, with a mid-IR image (8 µm), takenwith MSX, in the Cygnus X south region, we reveal the large-scale distribution of molecular gas and how it is influenced byUV radiation. Our prime target was the well-known bipolar neb-ula S106, which so far in the literature is widely considered as

being part of the “Great Cygnus Rift” at a distance of around600 pc.

In this paper, we show that several star-forming regions, in-cluding S106 and our Clouds A and B, are arranged on a large-scale circle of diameter ∼80 pc seen in both radio and mid-IRwave lengths. From their globular shape and the stratified dis-tribution of 8 µm and CO molecular line emission we concludethat the S106 molecular cloud in the south and two other star-forming regions in the north of the circle are directly shaped byUV radiation from inside the cavity.

The cavity was probably created by radiation and/or stel-lar wind compression from the massive stars of the OB (sub)-clusters in the Cygnus X region. To identify the main sources forUV radiation, we used the 2MASS survey to obtain a stellar den-sity map of Cygnus X south with stars brighter than magnitude10 in K-band. We identify the OB (sub)-clusters Cyg OB1,9, andNGC 6913 as the main sources for UV radiation. The distribu-tion of these bright and massive stars shows that the definition ofOB associations in terms of spatial extent and stellar content inthe Cygnus X south region needs to be revised.

The two clouds on the northern edge of the cavity are partof the “Cygnus X south” cloud complex (Schneider et al. 2006),which is clearly influenced by members of the Cyg OB1 andOB2 associations. From the large-scale morphology and the factthat the clouds in “Cygnus X south”, Clouds A and B, and S106are also tightly associated kinematically, we conclude that S106and Clouds A and B are associated with the large molecularcloud complex forming the “Cygnus X south region” and haveto be within the distance limits of the OB1/2 clusters, i.e., around1.7 kpc.

Due to this revised larger distance, the S106 molecular cloudis more massive (∼7600 M�) than previously thought. The spec-tral type of the exciting star (S106 IR) must be earlier than O9,which is indicated independently by a high [O III]/Hβ ratio in thevisible spectrum of S106 IR (Solf 1980).

Acknowledgements. This research made use of data products from theMidcourse Space Experiment. Processing of the data was funded by the BallisticMissile Defense Organization with additional support from NASA Office ofSpace Science. This research has also made use of the NASA/ IPAC InfraredScience Archive, which is operated by the Jet Propulsion Laboratory, CaliforniaInstitute of Technology, under contract with the National Aeronautics and SpaceAdministration.

References

Allen, D. A., & Penston, M. V. 1975, MNRAS, 172, 245Bally, J., & Scoville, N. Z. 1982, ApJ, 255, 497Bally, J., Yu, K. C., Rayner, J., & Zinnecker, H. 1998, AJ, 116, 1868Blaauw, A. 1964, ARA&A, 2, 213Boeche, C., Munari, U., Tomaselle, L., & Barbon, R. 2004, A&A, 415, 145Churchwell, E., & Bieging, J. H. 1982, ApJ, 258, 515Dame, T. M., Ungerechts, H., & Cohen, R. S., et al. 1987, ApJ, 322, 706Deharveng, L., Zavagno, A., & Caplan, J. 2005, A&A, 433, 565Dickel, H. R., & Wendker, H. J. 1978, A&A, 66, 289Dickel, H. R., Wendker, H. J., & Bierritz, J. H. 1969, A&A, 1, 270Downes, D., & Rinehart, R. 1966, ApJ, 144, 937Eiroa, C., Elsässer, H., & Lahulla, J. F. 1979, A&A, 74, 89Furuya, R. S., Kitamura, Y., Saito, M., et al. 1999, ApJ, 525, 821Ghosh, S. K., Kulkarni, V. K., Ojha, D. K., & Verma, R. P. 2003, Bull. Astron.

Soc. Ind., 31, 391Guillot, P. 2001, in From Darkness to Ligth, ed. T. Montmerle, & P. André, ASP

Conf. Ser., 243, 677Hayashi, S., Hasegawa, T., & Tanaka, M. 1990, ApJ, 354, 242Hodapp, K.-W., & Rayner, J. 1991, AJ, 102, 1108Humphreys, R. M. 1978, ApJS, 38, 309Ikhsanov, R. N. 1961, SvA, 4, 6Kramer, C., Stutzki, J., Röhrig, R., & Corneliussen, U. 1998, A&A, 329, 249

882 N. Schneider et al.: S106. III.

LeDuigou, J.-M., & Knödlseder, J. 2002, A&A, 392, 869Little, L., Kelly, M., Habing, R., & Millar, T. 1995, MNRAS, 277, 307Lucas, R., Le Squéren, A., Kazès, I., & Encrenaz, P. J. 1978, A&A, 66, 155Martins, F., & Plez, B. 2006, A&A, 457, 637Martins, F., Schaerer, D., & Hillier, D. J. 2005, A&A, 436, 1049Mel’Nik, A. M., & Efremov, Yu. N. 1995, Astron. Lett., 21, 10Maucherat, A. J. 1975, A&A, 45, 93Meynet, G., Maeder, A., Schaller, G., et al. 1994, A&AS, 103, 97Molinari, S., Brand, J., Cesarone, R., et al. 1998, A&A, 336, 339Motte, F., Bontemps, S., Schilke, P., Schneider, N., & Menten, K. 2007, A&A,

acceptedOasa, Y., Tamura, M., Nakajima, Y., et al. 2006, AJ, 131, 16080Persson, S. E., McGregor, P. J., & Campbell, B. 1988, ApJ, 326, 339Piepenbrink, A., & Wendker, H. J. 1988, A&A, 191, 313Pipher, J. L., Sharpless, S., Kerridge, S. J., et al. 1976, A&A, 51, 255Qin, S., Wang, J., Zhao, G., & Miller, M. 2005, Chin. J. Astron. Astrophys., 5, 1Rayner, J. 1994, Infrared Astronomy with Arrays (Kluwer Academic Press), 185Reipurth, B. 2007, Handbook of low mass star forming regions (ASP)Reifenstein, E. C., Wilson, T. L., & Burke, B. F. 1970, A&A, 4, 357Richer, J., Padman, R., Ward-Thompson, D., Hills, R. E., & Harris, A. I. 1993,

MNRAS, 262, 839Rieke, G. H., & Lebofsky, M. J. 1985, ApJ, 288, 618Ruprecht, J., Balazs, B. A., & White, R. E. 1981, Budapest: Akademiai KiadoSchaerer, D., de Koter, A., Schmutz, W., & Maeder, A. 1996, A&A, 310, 837

Schneider, N., & Brooks, K. J. 2004, PASA, 21, 290Schneider, N., Stutzki, J., Winnewisser, G., et al. 1998, A&A, 335, 1049Schneider, N., Simon, R., Kramer, C., Stutzki, J., & Bontemps, S. 2002, A&A,

384, 225Schneider, N., Simon, R., Kramer, C., et al. 2003, A&A, 406, 915Schneider, N., Bontemps, S., Simon, R., et al. 2006, A&A, 458, 855Sibille, F., Bergeat, J., Lunel, M., & Kander, R. 1975, A&A, 40, 441Simon, R., Jackson, J. M., Clemens, D. M., Bania, T. M., & Heyer, M. H. 2001,

ApJ, 551, 747Solf, J. 1980, A&A, 92, 51Spitzer, L. 1978, Physical processes is the ISM (New-York: Wiley Interscience)Stasinnska, G., & Schaerer, D. 1997, A&A, 322, 615Staude, H. J., Lenzen, R., Dyck, H. M., et al. 1982, ApJ, 255, 95Straizys, V., Kazlauskas, A., & Vansevicius, V., et al. 1993, Balt. Astron., 2, 171Stutzki, J., & Güsten, R. 1990, ApJ, 356, L63Stutzki, J., Ungerechts, H., & Winnewisser, G. 1982, A&A, 111, 201Uyaniker, B., Fürst, E., Reich, W., Aschenbach, B., & Wielebinski, R. 2001,

A&A, 371, 675Vallée, J. P., Fiege, J. D., & Jason, D. 2005, ApJ, 627, 263van den Ancker, M. E., Tielens, A. G. G. M., & Wesselius, P. R. 2000, A&A,

358, 1035Wang, J.-J., & Hu, J.-Y. 2000, A&A, 356, 118Wendker, H. J., Higgs, L. A., & Landecker, T. L. 1991, A&A, 251, 441White, G. J., Lefloch, B., Fridlund, C. V. M., et al. 1997, A&A, 323, 931